Skip to content

AI-enabled quality management

Quality documents an auditor trusts.AI that does the rest.

Doqumus replaces Word files on the quality manager's laptop with controlled revisions, real e-signatures, automatic supersession and an audit trail nobody can edit — including us. On top of that, an AI layer migrates your corpus, maps your clauses and analyses impact. It drafts and analyses. It never approves.

Built for ISO 9001 and API Spec Q1 manufacturers · Hosted in your region

SOP-12 · Rev C

Control of Documented Information

Effective
Effective
12 Aug 2026
Clauses
ISO 9001 §7.5 · API Q1 §4.3
  • R. Nair · AuthorPrepared · 04 Aug · 11:02 GST · step-up re-auth
  • S. Haddad · ReviewerReviewed · 05 Aug · 16:38 GST · step-up re-auth
  • A. Rahman · Quality ManagerApproved · 06 Aug · 09:41 GST · step-up re-auth
Content hashsha256:3f9a…c21b
Acknowledged34 / 36 assigned
ISO 9001:2015API Spec Q1Append-only audit trailData resident in your regionUnlimited users, never counted

The starting point

Word files, email approvals, and a folder called FINAL_v3

Most certified manufacturers run their quality system on documents that were never under control in the first place. It survives audits until it doesn't.

The shadow document

A procedure gets emailed, saved locally, edited, and printed. Now two versions are in use on the shop floor and nobody can say which one is current.

The approval you cannot prove

“Approved by email, 2023.” An auditor asks who approved it, when, against which revision, and what they were attesting to. The mailbox does not answer that.

The pre-audit scramble

Three weeks of evenings rebuilding a paper trail that should have been a by-product of doing the work. Every surveillance cycle. Every time.

The validated core

The boring part, done properly

Everything an auditor judges you on, built as a small system that changes slowly and is validated as a whole. This is the part that has to be right before anything clever is worth having.

Documents

Revisions with real effectivity

Numbering, revisions and effectivity dates as first-class concepts — not a status field someone remembers to change. When a revision becomes effective, its predecessor is withdrawn from every point of use in the same moment.
Signatures

E-signatures that mean something

Signing triggers step-up re-authentication. The signature binds identity, authentication time, the meaning of the signature and the hash of what was signed. Executed signatures are never altered or detached — corrections supersede, never erase.
Audit trail

A trail nobody can edit

Every regulated action is an append-only, hash-chained event. Monthly chain heads are anchored to write-once storage. No interface — administrative, support, or ours — can modify or delete an event.
Training

Read-and-acknowledge, automatically

When a revision goes effective, obligations are created for the roles that need it. Each acknowledgement binds a person to that revision at that time. “Trained on the current revision” is derived, never a flag somebody set.
Files

The approved bytes, provably

Every revision is stored by content hash in immutable storage. The file an auditor downloads in 2031 is provably the file that was signed in 2026.
Records

Everything relatable to everything

A typed link graph connects revisions, NCRs, CAPAs, training records, suppliers, equipment and audit findings — the substrate for traceability, evidence packs and change-impact analysis.

The AI layer

AI drafts and analyses. A human signs.

Most eQMS vendors bolt AI onto the record. We separated them architecturally: the AI layer reads everything and writes to exactly one place, and the only route from there into a controlled record is a human e-signature.

The AI layer reads everything and writes nothing except proposals. There is no other path in — not for an administrator, not for us.

Corpus migration

Thousands of legacy Word files, scanned PDFs and drawings, classified, de-duplicated and mapped into document families — including the ones nobody remembered.

Clause mapping & gap analysis

Your documents mapped against ISO 9001 and API Spec Q1 clause by clause, so the gaps are a list you can work through rather than a finding you receive.

Change-impact analysis

“SOP-12 is being revised — what else is affected?” Answered from the link graph: dependent procedures, forms, training obligations, open records.

Grounded search with citations

Ask a question in plain language, get an answer that cites the controlled revision and clause it came from. No citation, no answer.

Two audiences

A system judged twice, by people with different standards

Both have to be won. Losing either loses the quality system.

The quality manager

Compares every workflow against Word plus email — which is fast, familiar and free. So routing an approval has to take less effort than sending the email did, and the shop floor has to acknowledge a document on a tablet without training.

  • Approvals routed automatically by the approval matrix
  • Supersession handled by the system, not by memory
  • Training obligations raised the moment a revision goes effective
  • No licences to ration — everyone who needs the system has it

The auditor

Compares every record against ISO 9001 §7.5 and API Q1 document control, and is professionally sceptical about software that could have been edited after the fact. So the answer to “can this be changed?” has to be structural, not a policy statement.

  • Every regulated action an append-only, hash-chained event
  • Chain heads anchored to write-once storage each month
  • Signature meaning, identity and content hash bound together
  • The validation package regenerated with every release

Residency & isolation

Your quality records never leave your region

Hosted in Azure UAE North for Gulf customers. Every row of tenant data carries a tenant identifier enforced by the database itself; object storage is encrypted per tenant; embeddings, clusters and indexes are tenant-scoped.

Your content is never used to train models — not ours, not anyone's. Language-model calls are transient processing under contractual no-training terms, and a tenant can turn the AI layer off entirely. The core quality system works without it.

In-region at rest

Records, files and indexes stay in the customer region.

Isolated per tenant

Row-level security in the database, per-tenant storage keys.

No training on your data

Contractual, and structurally impossible across tenants.

Validation pack included

Requirements, tests and traceability, regenerated per release.

Pricing

We never count your users

Not your readers, not your authors, not your approvers. A quality system is only compliant when everyone who should be in it is in it, and pricing by headcount punishes exactly that. So people are not an input to the price at all — what you pay is set by how many sites you run and how many documents you keep under control.

Start by finding out what you actually have.

A free program you run on your own machine counts what is really on your shared drive: how many documents, how many are copies of each other, how much is scanned paper. It sends us nothing. Reading it together is the step after that, when you want it.